Transfers reduced boarding without safety signals

A structured transfer program increased appropriate community-hospital transfers and reduced emergency department boarding without observed transfer-related harm.
*Uncontrolled before-after quality-improvement study; Level 4 (OCEBM).

Citation

Xu O, Graven P, Burns B, Alday A, Merkel M, Clements J. Level-loading across a health system: a collaborative model for interhospital transfers. Journal of Hospital Medicine. 2026. doi:10.1002/jhm.70492

Background

Emergency department boarding delays care and limits access to specialty hospital beds. This study evaluated whether transferring stable admitted patients from an academic hospital to nearby community hospitals could improve patient flow.

Patients

Admitted emergency department patients at Oregon Health and Science University from September 2023 to December 2025. Patients needing specialized academic-center services, urgent unavailable procedures, or unsafe transport were not transfer candidates. Outcomes after emergency care excluded transfers to an outside hospital without shared records.

Intervention

Universal nurse screening for transfer eligibility, standardized patient communication, record-based tracking, hospitalist care for delayed transfers, dedicated ambulance coverage, and a transfer receiving unit.

Control

Pre-program usual transfer process and non-transferred admitted patients.

Outcome

Transfer volume, emergency department boarding time, inbound transfers, hospital stay, early deterioration, mortality, readmissions, and transfer-related adverse events.

Follow-up Period

28 months; clinical outcomes included 30-day mortality and readmissions.

Results

Outcome Key finding
Transfer volume Increased from 1.5 to 15.0 patients/week; rate ratio 10.3 (CI not reported).
Boarding time among transferred patients Fell from 6.5 to 2.9 hours across intervention cycles.
Adjusted boarding time 14.8 fewer hours for transferred patients (CI not reported).
Early deterioration 16.4 percentage-point lower adjusted probability (CI not reported); NNT about 7.
30-day readmission 3.7 percentage-point lower adjusted probability (CI not reported); NNT about 28.

No increase in adjusted length of stay or mortality was found. No transfer-related adverse events were identified. Results were analyzed at the encounter level, similar to intention-to-treat for a system workflow. The control reflected usual available care before implementation.

Limitations

Single health system, no randomization, and no concurrent control limit causal claims. Transferred patients were selected as more stable, so better outcomes may partly reflect lower illness severity. Some outcomes were unavailable for outside-hospital transfers, and confidence intervals were not reported.

Funding

Not reported; no conflicts declared.

Clinical Application

Hospitals with crowding may adopt structured transfer workflows for stable patients, but should monitor safety and avoid applying findings to higher-risk patients.